【作者】 李洋；

【导师】 张晓冬；

【作者基本信息】 北京交通大学 ， 电工理论与新技术， 2013， 博士

【摘要】 摘要：随着电气工程和其它学科的日益融合,特别是信息技术的快速发展,电气信息技术的广泛应用已经在许多方面深刻地改变了人们的日常生活方式,智能交通系统(ITS)就是其中的一项。作为ITS重要数据源的矢量电子地图的需求量也日益增加,而如何将栅格地图作为图源来生成矢量电子地图一直是研究的热门课题之一。地图匹配算法是智能交通系统中车辆导航的重要组成部分,而原始位置信息、数字地图以及高效匹配算法是提高地图匹配准确性的关键。本论文基于国家自然科学基金(兼容交通网络拓扑校正的地图匹配方法研究)项目,探索栅格交通地图道路网络的自动识别方法以及地图匹配的新算法,以便高效处理位置信号的动态误差,从而获得车辆位置与轨迹在数字地图上更准确的匹配。本论文主要研究成果和技术方法创新点如下：(1)利用噪声的自身特征和数学形态学,提出了基于像素点八方向外延特征的噪声完全再聚类的道路提取方法,消除了道路和区域上的文字噪声。(2)为了达到识别与提取完整的道路信息目标,研究了基于道路中心线违规判据与反馈循环处理的方法。仿真证明这种方法对规范化的城市交通地图道路图层识别效果显著。(3)把地图匹配的算法分为不同阶段：初始阶段匹配、路口区域匹配、车辆在匹配道路上的位置确定。按不同匹配阶段采用相应匹配算法,同时根据不同情况采取相应的补偿方式,减少了道路方向上的匹配误差。(4)根据概率理论和内积理论,构造了合适的不同权值函数,最后根据总的权值确定匹配参数。通过在Matlab平台上与目前现有算法进行仿真对比,该算法在不引入复杂运算的前提下,提高了匹配与显示准确度。更多还原

【Abstract】 ABSTRACT:With the increasing integration of the electrical engineering and other disciplines, especially the rapid development of information technology, our way of daily life has been profoundly changed by extensive application of the electrical information technology in many ways. Intelligent Transportation Systems (ITS) is one of them. As a basic part of ITS. vector maps are desired more and more, and it has been one hot topic how to make vector maps from grid maps as the map source. Map-matching algorithm is an important part of vehicle navigation in the intelligent transportation system. Based no the original location information and digital maps together with the efficient matching algorithm, it is key to improve the accuracy of map matching.This dissertation is based on the project of National Natural Science Fund "Research on map-matching method compatible with traffic network topology adjusting". It is method to explore grid map of road network transport of automatic identification, and the new map-matching algorithm can efficiently process the position signals with only dynamic error to insure the vehicle tracking with more accurate matching the digital map.Those thesis research and technological innovation are:(1) The use of noise with their own characteristics and mathematical morphology, the road are completely extracted from Re-clustering to the noise based on8direction of extension pixel in order to eliminate noise on the road and area.(2) Based on criteria of the illegal road centerline and methods of the feedback loop, the complete identification and extraction of the target road information are achieved. Simulation results show that the standardization of urban traffic on the road map layers recognition results are obvious.(3) The map-matching algorithm is divided into different stages:The matching of the initial stage, the matching of street corner district, determining the location of the vehicle on the matching road. The corresponding matching algorithm are adopted at different stages, and under different circumstances, the appropriate compensation are taken to reduce the matching error on the road directions.(4) Based on probability theory and the theory of inner product, a suitable function of different weights are constructed to determine the parameters of the final match with the total weight.On the Matlab platform of simulation, compared with the existing calculate way, the calculate way improve the accuracy of matching and display under the premise that don’t lead into a complicated operation.更多还原